A Phonocardiographic-Based Fiber-Optic Sensor and Adaptive Filtering System for Noninvasive Continuous Fetal Heart Rate Monitoring

Abstract

This paper focuses on the design, realization, and verification of a novel phonocardiographic- based fiber-optic sensor and adaptive signal processing system for noninvasive continuous fetal heart rate (fHR) monitoring. Our proposed system utilizes two Mach-Zehnder interferometeric sensors. Based on the analysis of real measurement data, we developed a simplified dynamic model for the generation and distribution of heart sounds throughout the human body. Building on this signal model, we then designed, implemented, and verified our adaptive signal processing system by implementing two stochastic gradient-based algorithms: the Least Mean Square Algorithm (LMS), and the Normalized Least Mean Square (NLMS) Algorithm. With this system we were able to extract the fHR information from high quality fetal phonocardiograms (fPCGs), filtered from abdominal maternal phonocardiograms (mPCGs) by performing fPCG signal peak detection. Common signal processing methods such as linear filtering, signal subtraction, and others could not be used for this purpose as fPCG and mPCG signals share overlapping frequency spectra. The performance of the adaptive system was evaluated by using both qualitative (gynecological studies) and quantitative measures such as: Signal-to-Noise Ratio-SNR, Root Mean Square Error-RMSE, Sensitivity-S+, and Positive Predictive Value-PPV.

Keywords: EMI-free; Least Mean Squares (LMS) algorithm; Normalized Least Mean Square (NLMS) algorithm; adaptive system; fetal heart rate (fHR); fetal heart sounds (fHS); fetal phonocardiography (fPCG); interferometer; maternal heart rate (mHR); maternal heart sounds (mHS).

Figure 1

Sample recordings of electrocardiogram (ECG) and phonocardiography (PCG) signals.

Figure 2

Noninvasive interferometric measurement probe.

Figure 3

Basic schematic diagram for our noninvasive PPG-based interferometric sensors and adaptive system for fHR monitoring.

Figure 4

Normalized Power Spectrum Density of data acquired from the test subject.

Figure 5

Basic scheme of an adaptive N-th order FIR filter with transversal structure and the LMS Algorithm.

Figure 6

Sample plots of real raw data acquired from the thoracic and abdominal sensors of two different test subjects. (a) volunteer No. 1; (b) volunteer No. 2.

Figure 7

Modelled raw signal measured by $I{S}_T$ in the abdominal region (mRR ∈ <12–16> rpm; mHR ∈ <65–85> bpm). (Left) 60 s; (Right) detail in the form of 2.3 s.

Figure 8

Modelled raw signal represented by $I{S}_A$ in the abdominal region (mRR ∈ <12–16> rpm; mHR ∈ <65–85> bpm, fHR ∈ <80–155> bpm, GA = 35 weeks, orientation of fetus: Right Occiput Posterior (ROP)). (Left) 60 s; (Right) detail in the form of 2.3 ss.

Figure 9

Recording of a reference signal for the adaptive system represented by $I{S}_T$ in the thoracic region.

Figure 10

Input signal of the adaptive system formed by a mixture of maternal heart rate (mHR) and fetal heart rate (fHR).

Figure 11

Output of the adaptive system when using (a) the Least Mean Square Algorithm (LMS) and (b) the Normalized Least Mean Square (NLMS) Algorithms.

Figure 12

Modelled reference (ideal) signal represented by $I{S}_A$ in the abdominal region without a maternal component (fHR ∈ <80–155> bpm, GA = 35 weeks, fetus position: Right Occiput Posterior (ROP)).

Figure 13

Comparison of reference and predicted time course of fHR (a) physiological case; (b) pathological case.

Figure 14

Bland-Altman statistics for reference and predicted values of fHR for (a) the LMS Algorithm; (b) The NLMS Algorithm.

Figure 15

Detailed analysis of output from the adaptive system using (a) the LMS; and (b) the NLMS Algorithms.